Dump your information architecture

Published on: August 7, 2006 by Paul Boag

Do you run a website made up of thousands of pages? If so, then you may want to consider listening to an excellent presentation by Russ Weakley. He proposes an interesting new approach to the information architecture of large sites.

I have written before about the unique challenges of coming up with navigational approaches for large sites. It is a subject particularly close to my heart as many of the sites I work on have huge numbers of pages. Sites of this size do not sit comfortably under traditional navigational models with information architectures that become very deep and convoluted.

The approach suggested by Russ in his presentation at Webstock is a radical departure and I can imagine a lot of hesitancy within larger organisations, however despite that, I still find it a fascinating concept.

Russ is proposing that website owners stop imposing navigational approaches on users of a site. Rather than creating an information architecture for a site, website owners should "let go control" and allow users to create their own paths to the information they require.

This is achieved through tagging. Instead of pages being grouped together in sections within a site's architecture, each page is essentially standalone. Visitors navigate to a page using tags inserted by the author of a page and via the search mechanism. By not imposing an artificial site architecture onto the site you avoid problems like cross linking between sections and navigational problems associated with deep sites.

It is a very hard concept to explain and so I suggest that if you are running a large site you take the time to download Russ' presentation.

Listen to Russ' presentation

The approach is not without its problems but it is something to seriously consider if the information architecture of your site is causing you problems.

I have to confess that although I find the concept very exciting I cannot see many websites adopting it. The concept of isolated pages not organised into a predefined architecture will simply seem too alien to many website owners. However, in a world where users "information forge" by dipping in and out of sites via search engines, it makes perfect sense to organise website pages in a non-hierarchal way.

I do see real potential in this approach, which is why I am intending to implement a similar approach on the upcoming relaunch of the boagworld website.

Comments

Comments are for the discussion of this post. If you have other questions / comments then post them to the forum or send me an email

Post by Richard Quick on August 7, 2006 11:09 AM

That's so spooky. I just finished listening to that presentation 10 minutes ago!

It was a really good podcast, and Russ is a great speaker.
Post by Jason Mayoff on August 7, 2006 2:32 PM

It makes a lot of sense to me, but I can see how some website owners might be a little taken aback by the suggestion. I wonder if they might be OK with a combination structure and tag system -- or would that just defeat the purpose. A tag-only system would certainly make our lives easier programming the site.
Post by Ross Johnson on August 8, 2006 5:09 AM

I see one major problem with this...

When the site starts up and has no tags... who is going to pick the tags? How does it get built to the point where tags actually present relevent information based on user input?

What if you have a huge site, that has yet to get lots of traffic? No one will tag your site, and those who do find your site will have poor navigation because 3-4 people picked the current tagging navigation.
Post by Paul Boag on August 8, 2006 8:42 AM

Listen to the his presentation Ross. The site is tagged by the site owner not the user.
Post by Lee on August 8, 2006 11:11 AM

I haven't listened to the presentation yet (downloading it now) but there are plenty of obvious concerns here.

If the site owner tags the documents, who says the tags make sense to the site's users? Even if you go through the usual IA and user-research methods to define your tags, the lack of a structure destroys relationships which would often naturally exist between documents.

If I wanted to find out how much my Council Tax was on my local authority website, I could reasonably expect to access the facility to pay it online from a related page. With tagging i'd have to search again, go through the results again...

As I say, haven't been able to listen to the podcast yet, but from my own experience and research a well-designed and visible taxonomy not only aids navigation, but also engenders trust on the part of the user.

Looking forward to hearing what Russ has to say though!
Post by Paul Boag on August 8, 2006 11:22 AM

There are certainly a lot of problems with it Lee. However related linking isnt one of them. Tagging would actually allow better related links because you could use tagging as an automated way of establishing which pages should sit together. Russ does a much better job at explaining it all than me so ill shut up now :)
Post by Ed on August 8, 2006 12:04 PM

It is worth pointing out that there are other formats of this talk available too (at least, I think it is the same talk, as I haven't finished downloading it yet!) (from the Webstock Recordings page):

MPEG-1 Video [533MB]

MP3 - Audio [25MB]

MP4 - Video [344MB]

Windows Media Stream (video)

Streaming Audio

Also there are Presentation slides
Post by Bob on August 8, 2006 2:18 PM

Paul, I'm interested in how this actually would work.

Would you have a collection of all tags listed somewhere? The home page?

What would happen if you had 200 000 000 000 (you get the idea) tags?

Even if you only listed the most popular wouldn't this leave some content inaccessible unless you knew what you were looking for and how to find it?

Just a few thoughts
Post by Lee on August 8, 2006 4:01 PM

Hi Paul, still haven't had time to listen to the full podcast but i've read the slides and it's interesting stuff.

I think the idea of using tags/metadata to allow users to search content is, in itself, a good one (and not a new one), but the idea of not providing a default taxonomy is where I disagree with it. I still feel that tags can cause confusion too, by associating data by tag but without context, but Russ does admit there are issues with this approach.

An interesting presentation though, and it certainly resonates with my experiences trying to develop IA for big sites. Often most of the difficulties lie with content authors and management who can't conceptualise their site any other way than the current way...
Post by Richard Quick on August 8, 2006 4:04 PM

Bob,

Your site's listed as: http://www.threedays.co.uk/

I know Christianity "borrowed" heavily from Judaism .. but don't you think it's a bit unethical to have a site that's so blatantly ripped off from this one, especially if you're trying to promote biblical values?

Any thoughts on whether "though shalt not steal" covered intellectual property?
Post by Paul Boag on August 8, 2006 4:07 PM

Ooo Richard, check before you rant mate! Not only is Bob a close friend of mine (and I gave him permission to use the design) but I am also a Christian and call down flaming wrath on anybody who takes the micky out of my fellow bible bashers!
Post by Richard Quick on August 8, 2006 4:16 PM

Woops!

Soz Bob - I take it all back.

;)

I wasn't taking the micky out of bible bashers, btw .. just ones that plagiarize.

As Bob's not one of those I'm a bit red faced at the moment.

I did look to see if it said anything on the site about you - but the fact it had a copyright message at the bottom from another web design company made me jump to wildly wrong conclusions.

Soz agin!

:)
Post by Bob on August 8, 2006 4:19 PM

*smile*
Always worth checking the about us page
Post by Paul Boag on August 8, 2006 4:21 PM

Ah there is nothing more enjoyable than watching somebody squirm (not a very Christian thing for me to say!).

No problem Richard. Nice to know you were ready to rush to my defense. Actually Bob is one of our developers. The other web design company is a little side business we let him and a few mates run. He is a great developer but crap designer (hence using our design).

I think what amused me most is that you thought so poorly of Bob that you thought he was thick enough to post his url on my site!

lol... you have really brightened up my afternoon Richard. Sorry to rib you ;)
Post by Richard Quick on August 8, 2006 4:31 PM

I think what amused me most is that you thought so poorly of Bob that you thought he was thick enough to post his url on my site!

Yeah, I did think that was a bit dumb.

:o)

Feel free to rib me any time - I'm used to it!
Post by Richard Conyard on August 8, 2006 6:21 PM

Getting back on thread it's an interesting concept, although something I think would work far better as a hybrid. Though still downloading.

Consider a standard top, left hand side navigation with related documents within right hand side navigation. In terms of tagging and meta lists perhaps the LGCL / APLAWS lists might make a good start, infact it's a model that would probably lend itself very well to local government sites that often have far more content than will fit nicely into an IA.
Post by Paul Boag on August 8, 2006 6:24 PM

I agree with the idea of using a hybrid approach. Thats actually what I am going to do with this site. The main emphasis of tags for me will be driving users to other articles on a similar topic.
Post by Richard Quick on August 8, 2006 6:35 PM

The only thing with a hybrid approcah is that it's not that different to what people have been doing for over 10 years.

For example, Amazon have "Customers who bought this item also bought.."

I agree that having tags is a good idea, but if all it is is something in the right-hand column it's not really an innovative idea, is it?

Flickr do exactly that, don't they?

I guess the idea of using tags in a large government or corporate site is rare.

Or is there something I'm missing?
Post by Richard Quick on August 8, 2006 8:31 PM

I've just listened to the Podcast again and I've come to a conclusion .. I think Russ has just invented the wiki (or at least something very similar).

- No navigation or traditional structure
- Tags
- Users can publish content and/or add comments
- Users control their own path through the content

It's an interesting idea to do a Wiki for a museum or a company, though.
Post by Sholom Sandalow on August 15, 2006 11:19 PM

The most obvious example of the diffeence between the 2 types of information architectures is Google and Yahoo.

Yahoo's was originally just a catalog of 'expert' defined categories. But before long it became obvious that that type of IA doesn't work effectively as a navigation tool, for a variety of reasons--primarily, because most categories are too purous to be defined precisely and consistently.

Part of the reason Google works so well is that the information architecture is completely user defined. Google's search algorithm places high priority on the term used in a link by a reffering site. So, when you do a search, the results are user defined. In a sense, the sum total of Google searchers are the experts defining categories.

However this IA strategy works well only if there is a (very) large user base, which can collectively define information spaces. Without it, the problems pointed out above will plague any web site that commits to this method.

A podcast for those who design, develop and run websites.

Latest shows

Have your say

Posts by tag

Dump your information architecture

Comments

Leave a comment

Additional Information

Interesting links

Supporting boagworld

Speaking and writing